Hello. I was a reader of Habr for a long time and finally I had the opportunity to publish my blog here. So, this is my debut on stage Habra. I work at Intel and do technical support for compilers. Accordingly, you have to deal with the compiler itself and about the compiler regions. For those who have never used Intel compilers, I have already talked a little about the basics of work here . Today I will talk a little about one of the components of the Intel Parallel Debugger Extension compiler (PDE). This extension is installed simultaneously with both Intel C ++ Compiler Pro and Intel Parallel Composer. It is possible to debug OpenMP * applications in both Linux and Windows. In order not to play the accordion, in the process of writing this post I conducted a small experiment. Various search engines asked about the Intel Parallel Debugger Extension. Below are the results:
As a result, the search showed that in the Russian-speaking community, I mostly wrote or told about PDE. Accordingly, I will tell (I repeat) about the main features only briefly:
Search for errors related to access to shared data;
Setting filters to ignore potential access errors to shared data;
Identify and stop when performing potentially reentrant functions;
Serialization of performance of parallel regions on the fly. In other words, the ability to run an application in sequential mode without rebuilding the application;
Window with SSE registers. Which allows you to view the contents of the SSE registers during debugging in various formats. And also change these values.
Now I want to ask you: “Have you heard or used PDE? What exactly did you like or dislike? What features need to be added? ”. That's all. Good luck, thank you.
Please refer to the Optimization Notice page for more details on performance and optimization in Intel software products.